2. MARCO TEÓRICO
2.1. Fundamentación teórica del problema
2.1.1. Fundamentación filosófica
2.1.7.2. Los métodos y su técnica de lanzamiento
2.1.7.2.5. Consideraciones generales sobre los lanzamientos al cesto
The nosological roots of affective disorders were planted in the notion that they were functional pathologies – that is the mood and behavioural disturbances occurred without obvious structural organic cause. Without known brain dysfunction, there was therefore little reason to expect patients to exhibit deficits on neuropsychological evaluation. However, almost fifty years ago Kiloh (1961) described a series of patients with psychiatric illnesses (predominantly depression) whose symptoms mimicked the cognitive changes seen in patients with dementia (Kiloh, 1962). As the cognitive symptoms tended to remit when the depression was treated, Kiloh termed it pseudo‐dementia. By the eighties and early nineties methodologically rigorous systematic studies of neuropsychological function emerged contrasting patients with major depression or bipolar disorder with healthy comparison subjects. Although rarely performing as poorly as patients with dementia (Christensen et al., 1997), depressed patients and patients with bipolar disorder showed deficits when acutely unwell with either depression or mania (Clark et al., 2001, Kurtz and Gerraty, 2009, Martinez‐Aran et al., 2004b, Zakzanis et al., 1998).
Focusing specifically on bipolar disorder, there is also evidence that cognitive impairment during an acute episode fails to remit fully when symptoms improve. There are in excess of 50 studies that have investigated at least one aspect of cognitive function in euthymic patients with bipolar disorder. Five independent research groups have published meta‐analyses which have shown very similar results – euthymic patients with bipolar disorder show deficits in several areas, but most prominently in verbal memory and aspects
of executive function (Arts et al., 2008, Bora et al., 2009, Kurtz and Gerraty, 2009, Robinson et al., 2006, Torres et al., 2007). Visual memory has been less commonly investigated, but in the most recent meta‐analysis showed an effect size similar to that for verbal memory. It is worth considering these findings a little further. COGNITIVE FUNCTION IN EUTHYMIA The following section is based on a systematic literature search (see Appendix 1 for full details). However, owing to the number of good quality reviews and meta‐analyses already available in this area (e.g. Arts et al., 2008, Bearden et al., 2001, Bora et al., 2009, Kurtz and Gerraty, 2009, Robinson et al., 2006, Torres et al., 2007) it was considered more appropriate to provide a selective review of the data most pertinent to the present study.
Table 1.1 below summarises the results of the meta‐analyses of cognitive function in euthymic patients with bipolar disorder. The meta‐analyses all draw on a common pool of studies and overall there is broad agreement between them. The difference between the highest and lowest reported effect sizes for any given measure vary between 0.01‐0.48. In general there is better agreement for the verbal and visual memory measures than the executive and attention measures. On closer inspection of the data, the differences in the attention measures most likely relate to methodological differences in how data from different measures was combined (particularly the continuous performance test measures, which were derived from several different but similar tasks).
The effect sizes are largest for measures of executive function, verbal list‐learning and visual memory. The lowest effect sizes are reported for visual copy, immediate verbal memory, and recognition memory. The only test index reported by each meta‐analysis to show a large effect size is total learning on the verbal memory list‐learning tasks. Within the executive functions, verbal fluency by letter and categories achieved on Table 1.1: Comparison of effect sizes (Cohen’s d) from the five meta‐analyses Test Index Robinson 2006 Torres
2007 Arts 2008 Bora 2009 Kurtz 2009 Executive Category fluency 1.09 ‐ 0.75 ‐ 0.75 Stroop 0.63 0.71 0.73 0.76 0.75 Trail Making Test B 0.78 0.55 0.73 0.86 0.73 Reverse Digit Span 0.98 0.54 1.02 0.75 0.65 WCST Perseverative Errors 0.76 ‐ 0.72 0.70 0.61 WCST Categories Achieved 0.62 0.69 0.49 0.66 0.54 Letter Fluency (FAS) 0.34 0.47 0.47 0.60 0.51 Verbal Memory A/CVLT total learning 0.90 0.81 ‐ 0.85 0.81 A/CVLT short delay recall 0.73 0.74 0.82 0.73 ‐ A/CVLT long delay recall 0.71 0.72 0.85 0.77 0.78 A/CVLT recognition ‐ 0.43 ‐ 0.44 ‐ Forward Digit Span 0.47 ‐ 0.37 0.37 0.41 WMS verbal memory immediate recall ‐ ‐ ‐ ‐ 0.63 WMS verbal memory delayed recall ‐ ‐ ‐ ‐ 0.92 Visuospatial ROCF copy ‐ ‐ 0.22 0.23 0.26 ROCF immediate recall ‐ ‐ 0.62 0.59 0.73 ROCF delayed recall ‐ ‐ ‐ ‐ 0.80 WAIS Block Design ‐ ‐ ‐ ‐ 0.55 Attention & Psychomotor CPT reaction time 0.60 0.62 ‐ ‐ ‐ CPT sensitivity 0.48 0.74 0.58 0.83 0.69 WAIS DSST 0.59 0.79 0.84 0.75 0.66 Trail Making Test A 0.52 0.60 0.58 0.69 0.65
Key: White Small effect size (0.2≤d<0.5) Medium effect size (0.5≤d<0.8) Large effect size (d≥0.8)
WCST, Wisconsin Card Sorting Test; A/CVLT, Rey/California Auditory Verbal Learning Test; WMS, Wechsler Memory Scale; ROCF, Rey‐Osterrieth Complex Figure; WAIS, Wechsler Adult Intelligence Scale; CPT, Continuous Performance Test; DSST Digit Symbol Substitution Test
the Wisconsin Card Sorting Test consistently show lower effect sizes (generally d<0.65) than the other executive functions. The remaining measures show a very similar degree of impairment (generally 0.7<d<0.8).
The evidence indicates relatively broad impairment in patients with bipolar disorder, with the most pronounced impairment in verbal memory, visual memory, and several aspects of executive function (e.g. category fluency, inhibition, set‐shifting and mental manipulation). Discounting a few extreme results, the largest deficits tend to be in the range 0.7<d<0.85.
COGNITIVE DYSFUNCTION AS AN ENDOPHENOTYPE
There has been much debate about whether cognitive dysfunction in bipolar disorder could be an endophenotype. The observed behaviours and symptoms that form the phenotype of bipolar disorder show significant heterogeneity both between individuals and within an individual over time. Some of this heterogeneity is inherent in the criterion‐based diagnostic system used to define the limits of what comprises the illness. Two different individuals with ostensibly the same diagnosis can have remarkably different presentations, to the extent it may be feasible to question whether they have the same underlying illness. The difficulty for research, particularly that which is focused on the underlying genetics, is identifying aspects or traits of the illness that lie somewhere between the genotype and the phenotype – a so‐called endophenotype – that can be used to identify subgroups of individuals more likely to share commonalities in their illness pattern or aetiology.
To satisfy the criteria expected of an endophenotype, cognitive dysfunction needs to be demonstrated to be associated with the illness, independent of mood state, heritable, and present in relatives of those with the disorder more commonly than observed in the background population (Hasler et al., 2006). It performs variably on these four criteria.
Firstly, with respect to association with the illness, although there is a large body of evidence (described above) showing that patients with bipolar disorder perform more poorly than healthy controls on neuropsychological assessment, there is no convincing evidence that the deficits observed are specific to bipolar disorder. Patients with schizophrenia (Barch, 2009), obsessive compulsive disorder (Aigner et al., 2007, Segalas et al., 2008), major depression (Austin et al., 1999, Paradiso et al., 1997), or personality disorders (Dinn et al., 2004, Monarch et al., 2004, Ruocco, 2005) also show impairments on many of the same cognitive measures. This may reflect the broad‐based sensitivity of many of the neuropsychological measures commonly used, or that similar cognitive processes are affected in a variety of psychiatric pathologies.
Secondly, the deficits are independent of mood state, in that there are a core set of impairments noted in euthymic patients, but symptomatic patients tend to show a more extensive range of deficits than those who are well. There is also some debate about the level of subsyndromal symptoms that are relevant in terms of their impact on neuropsychological function. Studies of euthymic patients rarely recruit individuals with no symptoms at all, but rather those with very low levels of symptoms that – in the context of their illness – reflects a state of relative wellness. However, most studies report significant differences between patients and controls on their symptom levels, even in patients whose scores are nonetheless very low. Although statistical control for symptom levels rarely renders all neuropsychological differences statistically non‐significant, a question remains whether statistical techniques can reasonably correct for something that is fundamentally different between the groups rather than a difference that has simply arisen by the misfortune of chance – the former representing circumstances under which covariance techniques are inappropriate (Strauss and Allred, 1987). Analysis of covariance (ANCOVA) was originally designed to increase power to detect between‐group differences by reducing within‐group variance that had arisen due to chance differences between groups. ANCOVA
therefore assumes random allocation to groups. Patients and controls are not allocated randomly to groups and measures such as symptom scores do not vary randomly between the groups. Instead they tend to vary systematically and are in fact intrinsically related to group membership. Using symptom scores as covariates in these circumstances is generally inappropriate and may under‐correct for their ‘true’ effects, that is statistically significant differences may remain evident when they ought not to.
Thirdly, with regard to heritability, although there is evidence that cognitive function in general shows a degree of heritability, there is only a single study investigating heritability of cognitive function specifically in bipolar disorder (Antila et al., 2007). This study reported a statistically significant degree of heritability for psychomotor speed, working memory, and executive function, but no significant heritability for verbal learning. This is an interesting finding given that verbal learning is one of the areas of greatest impairment in bipolar patients, yet it does not show evidence of heritability. In a similar vein, Szoke et al reported significant familial resemblance between bipolar patients and their first degree relatives for executive function and psychomotor speed measures (no other domains were assessed in this study) (Szoke et al., 2006). Taken together these results provide evidence that cognitive function in bipolar disorder is heritable and some of the areas showing heritability are those where impairment has been found in patients. However, the studies to date have only included patient samples where bipolar disorder runs in the family and have comprised probands and their first degree relatives (with only a small subsample of second degree relatives in the study by Antila et al). Study samples with a greater range of relatedness which also include cases with no prior family history of the illness are necessary to build on these initial findings. Also, one of the most robust areas of cognitive impairment in bipolar disorder – verbal memory – did not show significant heritability. If confirmed in further studies, this anomaly would require investigation and may hint at the possibility that there is more than one process leading to cognitive impairment in bipolar disorder.
Finally, there is evidence that unaffected first degree relatives of patients with bipolar disorder show subtle cognitive impairments when contrasted with healthy individuals with a benign family psychiatric history (Ferrier et al., 2004, Keri et al., 2001, Sobczak et al., 2002, Zalla et al., 2004). The deficits are smaller in magnitude than those found in patient samples and are generally restricted to verbal memory and executive function. However, some studies have investigated individuals below the mean age of onset of bipolar disorder and in any study including high risk groups some of the participants may yet go on to develop the condition. Without a substantial period of follow‐up it is therefore not possible to ascertain whether in some of these individuals, the cognitive deficits are a prodromal sign of illness.
All in all, on present evidence alone, there is not strong support for cognitive dysfunction as an endophenotype in bipolar disorder. However, the state of the evidence itself is somewhat lacking. There is a growing trend for studies to divide patient samples on potentially relevant characteristics, such as history of psychosis, substance misuse, level of social functioning and other severity of illness indicators, which is beginning to provide more detail about factors relevant for cognitive functioning in bipolar disorder (Ferrier et al., 1999, Glahn et al., 2007, Martinez‐Aran et al., 2008, Van Gorp et al., 1998). Some studies are beginning to delve into greater depth to understand the core processes behind the reported deficits, using novel tasks or novel analysis methods to attempt to tease apart the multiple different processes which contribute to any single neuropsychological task (Glahn et al., 2006, Thompson et al., 2007, Thompson et al., 2006). Although in its infancy, this line of inquiry may ultimately result in the development of more specific test batteries that are better able to indicate the major factors driving impaired performance.
AETIOLOGICAL ISSUES
One of the major issues that remains largely unknown is when cognitive function develops in bipolar disorder. Does it pre‐date illness and therefore potentially hold insights into the underlying pathology, or does it develop after mood symptoms and shed light on the consequences of the disorder or its treatment? The gold standard evidence – prospective longitudinal data including the pre‐illness period – is rare and so inferences have to be made from a variety of study designs.
LONGITUDINAL STUDIES
PROSPECTIVE STUDY
There has been one longitudinal prospective study of individuals at high risk for mood disorder, which reported that two thirds of individuals who met criteria for bipolar disorder by early adulthood had shown impaired performance on the Wisconsin Card Sorting Test (WCST) when assessed in adolescence (Meyer et al., 2004). In contrast, only one fifth of those who developed unipolar major depression showed impaired WCST performance, which was comparable to the rate found in those who did not develop any major mood disorder. However, this study involved only a small number of participants who ultimately developed bipolar disorder (n=9), two of whom already had bipolar disorder in adolescence. A larger study is necessary to confirm this finding.
RETROSPECTIVE STUDIES
Three retrospective studies conducted in countries with comprehensive population registers used cohorts of conscripts to examine the association between cognitive performance at conscription and subsequent development of psychiatric disorder (Reichenberg et al., 2002, Tiihonen et al., 2005, Zammit et al., 2004). Zammit et al (2004) reported no relationship between IQ and bipolar disorder (whereas low IQ was associated
with later development of schizophrenia). Similarly, Reichenberg et al (2002) reported no relationship between cognitive performance and later diagnosis of nonpsychotic bipolar disorder in a cohort of Israeli conscripts. In contrast, Tiihonen et al (2005) reported that poor performance on the visuospatial reasoning subtest of the Finnish Defence Forces Basic Ability Test was associated with later hospitalisation for bipolar disorder and schizophrenia. However, bipolar disorder was associated with spared or superior performance on the mathematics subtest, whereas schizophrenia was associated with poor performance on this subtest. Performance on the verbal reasoning subtest did not (independently) predict subsequent psychiatric illness. Neither the Zammit et al (2004) study nor the Reichenberg et al (2002) study examined performance subscales of the IQ tests. The relationship reported in the Tiihonen et al (2005) study was specifically with hospitalization for bipolar disorder rather than simply diagnosis of bipolar disorder, which diverges from the other two studies. This may account for the apparent differences between them; those who go on to have episodes severe enough to require hospitalization may be more likely to show cognitive difficulties.
These retrospective case‐register studies suffer several limitations ‐ notably the tests used are usually generic tests suitable for mass‐administration or for administration with relatively little training; the tests are aimed to identify strengths and weaknesses for channeling recruits into their most appropriate role, they are not designed to detect cognitive impairment and their psychometric properties for that purpose remain largely unexplored; psychiatric symptoms are not thoroughly assessed at the time of testing (beyond medical checks to establish sufficient health for serving in the armed forces) and some individuals may already be showing symptoms or prodromal signs of illness that are not acute or severe enough to be detected. Nonetheless, the naturalistic nature of the data and the large sample sizes make these studies a rich source of information, and have indicated that in generic assessment any prodrome to bipolar disorder is at best subtle and
for the majority of individuals does not represent markedly anomalous cognitive function before the onset of symptoms.
POST‐ONSET LONGITUDINAL STUDIES
After onset of illness, is there any evidence that cognitive performance declines over time? There are a small number of test‐retest studies that begin to address whether cognitive function shows deterioration over the course of the illness (Balanza‐Martinez et al., 2005, Gildengers et al., 2009, Mur et al., 2008). The longest of these ‐ conducted over a three‐year interval ‐ reported evidence of cognitive impairment at both baseline and follow‐ up, but no evidence of significant deterioration between assessments (Balanza‐Martinez et al., 2005). In a similar study conducted over two years, again patients showed evidence of impairment but the authors reported no significant deterioration between assessments (Mur et al., 2008). However, there was a significant interaction between group and time for verbal learning, such that patients with bipolar disorder worsened slightly over time whereas controls showed a slight improvement. Given that neither change was significant in itself, the authors dismissed the finding and focused on the persistent and stable executive function impairment seen in the patients. However, the differential trajectory of memory function between patients and controls may be of importance. To clarify whether this was indicative of a difference between the groups, further analysis comparing performance at the endpoint controlling for baseline performance, or analysis of predicted scores at endpoint accounting for baseline performance and practice effects would have been informative. This relatively subtle suggestion of differential memory function in bipolar patients may be indicative of a genuine effect that would be more evident in a larger sample followed‐up for a longer period of time.
A few studies have focused specifically on older patients with bipolar disorder. Depp et al (2008) reported no difference in the trajectory of cognitive change in middle‐aged and
older bipolar patients compared to healthy controls over 1‐3 years of follow‐up, but they did note greater within‐subject variation in the participants with bipolar disorder than either healthy controls or a comparison group with Schizophrenia (Depp et al., 2008). However, participants in this study were not necessarily euthymic or in the same mood state at the two different testing points (in contrast to some other similar studies). The baseline demographic data indicate that the bipolar patients were experiencing a higher mean level of depressive symptoms with a larger variance than the other groups which may account for the greater variation in performance. In a sample of older patients with bipolar disorder (all over 50 years of age), Gildengers et al (2009) showed evidence of more rapid cognitive decline over the period of three years compared to healthy controls. Using a generic dementia‐screening battery administered annually, bipolar patients showed impairment at all time points, and whereas controls showed only a slight decline in performance over the three years, the patient sample showed gradual deterioration year on year. By the end of the study eight patients scored below the cut‐off for dementia, in contrast to none of the participants in the control group. This study suffered large attrition, with less than fifty percent of the initial sample remaining in the study by the finial assessment point, which may have influenced the results especially as attrition was higher amongst the relatively younger participants. Likewise, although the average age of illness onset for the whole sample was consistent with that usually seen for the disorder (late‐twenties/early thirties), the range indicated there was at least one individual with very late onset (>70 years). Late